Cement and Concrete Research 35 (2005) 609 – 613

The use of thermal analysis in assessing the effect of

temperature on a cement paste
Lucia Alarcon-Ruiza,*, Gerard Platretb, Etienne Massieub, Alain Ehrlachera
a
Laboratoire d’Analyse des Matériaux et Identification, Institut Navier, 6 et 8 Av. Blaise Pascal-Cité Descartes-Champs sur marne,
77455 Marne la vallée CEDEX 2, France
b
PCM, Laboratoire Central des Ponts et Chaussées, 58 boulevard Lefébvre 75732 Paris cedex 15, France
Received 27 January 2003; accepted 8 June 2004

Abstract

Upon heating, the cement paste undergoes a continuous sequence of more or less irreversible decomposition reactions. This paper
reports studies on a cement paste fired to various temperature regimes up to 800 jC in steps of 100 jC for a constant period of 24 h. This
work has been carried out using thermal analysis technique to study the effect of temperature in the mineralogical composition of cement
hydration products. The thermal decomposition of the cement paste is analysed with the thermogravimetric analysis (TGA) and the
derivative thermogravimetric analysis (DTG) curves. Such techniques can be used to determine fire conditions and the consequent
deterioration expected in the cement paste. Therefore, the aim of this work is to have a better knowledge of the reactions that take place in
a cement paste during a fire and thus to be able to determine the temperature history of concrete after a fire exposure. The results show
that even if the dehydroxylation reaction is reversible, the portlandite formed during the cooling has an onset temperature of
decomposition lower than the original portlandite and can thus be considered as a tracer for determining the temperature history of
concrete after a fire exposure.
D 2004 Elsevier Ltd. All rights reserved.

Keywords: Thermal analysis; Cement; C-S-H; Portlandite; Fire

1. Introduction concrete could have been exposed during an accidental fire

[4]. Raina et al. [5] and Handoo et al. [6] have also reported
Thermoanalytical techniques, including differential ther- the application of DTA/TGA technique for assessing fire-
mal analysis and derivative thermogravimetry, have all been damaged concretes. Recently, Handoo et al. [7] have shown
used successfully [1– 3], over the years, for determining the that the decrease in portlandite content with increase in
temperature history of concrete after fire exposure. temperatures can be used for assessing the condition of
The concrete is mainly composed of cement paste and building elements subjected to accidental fires.
aggregates. Harmathy [4] explains that the investigations with As the cement paste is the component of the concrete that
thermoanalytical techniques make use of the fact that, upon makes all these investigations possible and as the presence of
heating, the cement paste undergoes a continuous sequence of aggregates tends to mask certain features of the thermogravi-
more or less irreversible decomposition reactions. Therefore, metric curves [4], the present work analyses the thermal
once heated piece of the specimen will behave for some time effects on a cement paste by studying the thermogravimetric
as a stable material up to the temperature of heating. The analysis (TGA) and the derivative thermogravimetric (DTG)
author has demonstrated the use of the differential thermog- curves.
ravimetric technique for gauging the temperature to which

2. Theoretical considerations
* Corresponding author. Tel.: +33-1-64-15-37-01; fax: +33-1-64-15-
37-41. Hydrated cement paste is composed of four major com-
E-mail address: alarcon@lami.enpc.fr (L. Alarcon-Ruiz). pounds: tricalcium silicate (C3S), dicalcium silicate (C2S),

0008-8846/$ – see front matter D 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2004.06.015
610 L. Alarcon-Ruiz et al. / Cement and Concrete Research 35 (2005) 609–613

tricalcium aluminate (C3A) and tetracalcium aluminoferrite Table 2

Physical characteristics of cement
(C4AF). The most important products of the hydration
reactions are the calcium silicate hydrate (C-S-H) and the Blaine’s (m2/kg) Density (g/cm3) Compressive strength (MPa)
portlandite, also called calcium hydroxyde, (CH) [8]. 1 day 2 days 7 days 28 days
Different authors have described the reactions that occur 351.5 3.02 10.0 21.5 37.0 48.5
with an increase of temperature in cement paste and concrete:

 30 – 105 jC: the evaporable water and a part of the bound constant period of 24 h. The samples were then left in the
water escapes. It is generally considered that the eva- furnace until they cool to ambient temperature and were
porable water is completely eliminated at 120 jC [9] stored afterwards in a dessicator.
 110– 170 jC: the decomposition of gypsum (with a The test samples were cut from the inner core of these
double endothermal reaction) [9,10], the decomposition cylinders. The amount of material to be subjected to TGA
of ettringite [11] and the loss of water from part of the was crushed and ground until a grain size of 80 Am is
carboaluminate hydrates [13] take place obtained. The apparatus used in the thermogravimetric
 180 –300 jC: the loss of bound water from the de- studies was a NETZSCH STA 409. The temperature of
composition of the C-S-H and carboaluminate hydrates the furnace was programmed to rise at constant heating
undergoes [9,12,14] rate of 5 jC/min up to 1150 jC. The tests were performed
 450 –550 jC: dehydroxylation of the portlandite (calci- under an air flow of 80 ml/min. The amount of material
um hydroxyde) [9,10] used for the tests was weight with a balance of F 0.1 mg
 700– 900 jC: decarbonation of calcium carbonate [9,15] of accuracy and was ranging between 220 and 226 mg.

To analyse the thermal effects in the cement paste, we have

studied the DTG and the TGA. The DTG curves have marked 4. Results and discussion
inflexion points and plateaus, features that are not present in
the TGA curves. Therefore, DTG curve better marks all The results of the thermogravimetric tests obtained at a
different processes than its precursor, TGA. A change in rate temperature of 5 jC/min are summarized in Fig. 1.
TGA slope is reflected as a marked peak in DTG [16]. The reference specimen corresponds to an unfired
sample of cement paste. It can be observed that the curve
of the reference specimen shows three rapid weight loss.
3. Experimental procedure The first weight loss, located between 100 and 200 jC, is
the result of dehydration reactions of several hydrates (C-
The cement used for this study was taken from Gaurain S-H, carboaluminates, ettringite, etc.). This loss of water
in Belgium. The chemical analysis of the CEM II/B indi- is mainly due to the dehydration of the C-S-H. Conse-
cating the presence of main constituent oxides and the main quently, we will include the different hydrates under the
components of the cement are given in Table 1. Some of the annotation C-S-H.
cement paste physical characteristics are listed in Table 2. The second major weight loss, observed at 450 – 500 jC,
The samples to be investigated were originally for other corresponds to the dehydroxylation of portlandite, another
studies [17 – 19]. The cement paste was made with a water/ hydration product.
cement ratio of 0.33. The cement paste cylinders (diameter The third loss of weight appears at 750 jC and corre-
of 3 cm and height of 8 cm) were stored for at least 28 days sponds to the decarbonation of calcium carbonate coming
under water at a temperature of 22 F 2 jC while waiting to from the clinker and the filler.
proceed to the heat treatment. Let us consider the behaviour of the heat-treated
The water-cured cylinders were heated in a PROLABO samples. Fig. 1 shows that the second weight loss, cor-
Volca MC 25 furnace at a heating rate of 100 F 10 jC responding to the dehydroxylation reaction, remains even
every 90 min up to various temperature regimes for a for samples fired beyond 500 jC. If we consider that, for
a previous heat treatment beyond 600 jC during 24 h,
Table1
the dehydroxylation reaction is completed in the sample,
Oxide constituents and main components (%) of the cement then it can be deduced that this reaction is reversible.
Oxide constituents (%)
Nevertheless, it is of notice that the beginning of the
weight loss seems to shift towards lower temperatures
CaO SiO2 Al2O3 Fe2O3 MgO SO3 TiO2 K2O Na2O P2O5
when the temperature of the previous heat treatment
59.00 19.45 4.80 1.75 3.10 2.95 0.30 0.70 0.19 0.23 exceeds 500 jC.
Main components Moreover, considering the decarbonation reaction, the
results show that the samples fired to temperatures above
Clinker Slag Calcareaus Gypsum
600 jC do not present the third weight loss. Thus, it can be
73 14 13 3.5 deduced that this reaction is irreversible.
 L. Alarcon-Ruiz et al. / Cement and Concrete Research 35 (2005) 609–613 611

Fig. 1. TGA curves of cement paste fired to various temperature regimes.

The weight loss for each sample in the three different A comparison between the curve of reference to the
reactions are summarized in the Table 3. curves representing heat-treated cement paste samples re-
These values are reported in the Fig. 2. We can observe veals that, as expected, above the temperature of previous
the evolution of each reaction according to the previous heat heating, the various curves are close to the thermogravi-
treatment. metric curve of the reference specimen. The peak of the
The dehydration reaction of the hydrates decreases grad- dehydration of the C-S-H reaction decreases with the
ually and can be considered as an irreversible reaction. increase of the previous heat treatment and disappears for
Therefore, the dehydration reaction can be used as an a heat treatment superior to 300 jC. These results are in
indicator to determine the temperature history of concrete agreement with Refs. [9,14].
after a fire exposure. We can observe that the peak of decarbonation of
The dehydroxylation and decarbonation reactions are calcium carbonate disappears completely for a heat treat-
correlated. We can observe that for the samples fired up ment of 800 jC as expected [9]. This reaction can therefore
to 700 and 800 jC, the increase of portlandite is corre- be used as a marker for determining the temperature history
lated to the decrease of the decarbonation reaction. This is of concrete exposed to a fire regime.
mainly due to the recrystallization of the portlandite The second peak corresponding to the dehydroxylation
during the cooling in the furnace after the heat treatment. of portlandite does not disappear completely even for
These reactions are studied more carefully with the DTG previous heat treatment of 800 jC. This is due to a
results. recrystallization of the amorphous part of portlandite [20]
The DTG curves are represented in Fig. 3. By analysing after the heat treatment during the cooling in the furnace.
the different peaks, it is possible to obtain several parame- This shows the importance to make the tests very quickly
ters as the temperature at different decomposition degrees after a fire. Harmathy [4] wrote that the maximum temper-
and the mass loss in each process. atures attained at various locations in a concrete construc-
A simple view to the DTG curve of reference shows that tion during a fire exposure can be determined if the samples
the DTG better marks all different process than the TGA. of the concrete can be procured within 1 or 2 days of the
We can observe a smaller peak in the lap of the dehydra- fire. A better study of the portlandite dehydroxylation is
tation of C-S-H peak at 170 jC. The peak probably done in Fig. 4.
corresponds to the monocarboaluminates’ decomposition. In Fig. 4, we have reported the onset temperature of the
This reaction is very difficult to see in the TGA curves as it peak for the dehydroxylation reaction. One can observe
is just a slightly change of slope. three main ranges for the onset temperature. For a tem-

Table 3
Loss of weight (%)
Reaction Ref. 100 jC 200 jC 300 jC 400 jC 500 jC 600 jC 700 jC 800 jC
(1) 79.83 54.09 41.82 20.13 10.68 12.16 8.22 11.21 0.00
(2) 9.05 25.51 28.25 44.61 37.91 26.56 27.59 57.01 88.76
(3) 11.12 20.41 29.92 35.25 51.41 61.28 64.19 31.78 11.24
Reaction (1): dehydration reactions mainly due to the loss of water from C-S-H. Reaction (2): dehydroxylation of porlandite. Reaction (3): decarbonation of
calcium carbonate.
612 L. Alarcon-Ruiz et al. / Cement and Concrete Research 35 (2005) 609–613

Fig. 2. Loss of weight (%) for the dehydration reaction (C-S-H), the dehydroxylation reaction (portlandite) and the decarbonation reaction (calcium carbonates)
versus previous temperature treatment.

Fig. 3. DTG curves of cement paste fired to various temperature regimes.

Fig. 4. Onset temperature of the peak for the dehydroxylation reaction versus previous temperature treatment.
 L. Alarcon-Ruiz et al. / Cement and Concrete Research 35 (2005) 609–613 613

perature treatment below 400 jC, the onset temperature is References

almost constant and close to 460 jC. If we consider now
the previous thermal treatment above 600 jC, we can [1] H.G. Midgley, The use of thermal analysis methods in assessing the
observe that the onset temperature range is also almost quality of high alumina cement concrete, J. Therm. Anal. 13 (1978)
515 – 524.
constant but the values are close to 386 jC. Finally, the [2] V. Baroghel-Bouny, Caractérisation microstructurale et hydrique des
samples fired up to 400 or 500 jC have an onset pâtes de ciment et des bétons ordinaires et à très hautes performances,
temperature value intermediary. This means that the recrys- PhD thesis, Ecole Nationale des Ponts et Chaussées, 1994.
tallized portlandite is not as thermically stable as the [3] C. Krzys, Analyse de trois méthodes de détermination du degré d’hy-
portlandite of origin and can be identified by the onset dratation du béton (Training report), Service Physico-chimique du
Laboratoire Centrale des Ponts et Chaussées, Paris, 1999.
temperature of the peak. [4] T.Z. Harmathy, Determining the temperature history of concrete Insti-
The onset temperature of the portlandite dehydroxyla- tut Naviere constructions following fire exposure, ACI J. 65 (11)
tion reaction peak can be used as a marker for determin- (1968) 959 – 964.
ing the temperature history of concrete after a fire [5] S.J. Raina, V.N. Vishwanathan, S.N. Ghosh, Instrumental techniques
exposure. for investigation of damaged concrete, Indian Concr. J. 52 (1978)
147 – 149.
[6] S.K. Handoo, S. Agarwal, S.K. Agarwal, S.C. Ahluwalia, Effect of
temperature on the physico-chemical characteristics of hardened con-
5. Conclusion crete, in: H. Justnes (Ed.), 10th International Congress of Chemistry of
Cement, Gothenburg, Sweden, June 2 – 6 4IV 067 (1997) 4.
[7] S.K. Handoo, S. Agarwal, S.K. Agarwal, Physicochemical, mineral-
In this paper, a cement paste fired to various temperature
ogical, and morphological characteristics of concrete exposed to ele-
regimes up to 800 jC has been studied using thermal vated temperatures, Cem. Concr. Res. 32 (7) (2002) 1009 – 1018.
analysis technique. The investigations with thermal analysis [8] W. Sha, E.A. O’Neill, Z. Guo, Differential scanning calorimetry
make use of the fact that during the heating, the cement study of ordinary Portland cement, Cem. Concr. Res. 29 (9) (1999)
paste undergoes a continuous sequence of more or less 1487 – 1489.
irreversible decomposition reaction. The cement paste is [9] A. Noumowé, Effet des hautes températures (20 jC – 600 jC) sur le
béton, PhD thesis, Institut National des Sciences Appliquées, 1995.
therefore the component of the concrete which makes these [10] G. Platret, Suivi de l’hydratation du ciment et de l’évolution des
investigations possible. phases solides dans les bétons par analyse thermique. Caractéristiques
The accomplishments and conclusions of the study are microstructurales et propriétés relatives à la durabilité des bétons.
summarized below: Méthodes de mesure et d’essai de laboratoire, Méthodes d’essai No.
58, Laboratoire Central des Ponts et Chaussées, Février, 2002.
[11] Q. Zhou, F.P. Glasser, Thermal stability and decomposition mecha-
 The TGA curves have shown that the dehydration and nisms of ettringite at < 120j, Cem. Concr. Res. 31 (9) (2001)
the decarbonation reactions are irreversible in the cement 1333 – 1339.
paste and can be used as tracers for determining the [12] N. Richard, Structure et propriétés élastiques des phases cimentiéres à
temperature history of concrete after a fire exposure. base de monoaluminate de calcium, PhD thesis, Univ. Paris VI, 1999.
 The dehydroxylation reaction of the portlandite is [13] E. Nonnet, N. Lequeux, P. Boch, Elastic properties of high alumina
cement castables from room temperature to 1600 jC, J. Eur. Ceram.
reversible. This hydration product seems to rehydrate Soc. 19 (1999) 1575 – 1583.
rapidly after a heat treatment. [14] G.A. Khoury, Compressive strength of concrete at high temperatures:
 The derivative thermogravimetric curves give the onset a reassessment, Mag. Concr. Res. 44 (161) (1992) 291 – 309.
temperature of the reactions. The onset temperature of [15] P.E. Grattan-Bellew, Microstructural investigation of deteriorated
the dehydroxylation reaction decreases (from 460 to Portland cement concretes, Constr. Build. Mater. 10 (1) (1996) 3 – 16.
[16] F. Guirado, S. Gali, J.S. Chinchon, Thermal decomposition of hydrat-
386 jC) for a previous heat treatment above 500 jC. ed alumina (CAH10), Cem. Concr. Res. 28 (3) (1998) 381 – 390.
[17] L. Alarcon-Ruiz, Analyse de l’évolution de la microstructure de la
In prospect, other studies about the onset temperature pâte de ciment sous chargements thermiques, PhD thesis, Ecole Natio-
of the peak of the dehydroxylation reaction in addition nale des Ponts et Chaussées, 2003.
to the portlandite content in the cement paste must be [18] L. Alarcon-Ruiz, D. Pasquero, A. Ehrlacher, Modélisation de l’évolu-
tion de la porosité avec la déshydratation des pâtes de ciment, . Na-
done. tional Conference, Matériaux 2002, Tours, France, Octobre 21 – 25.
Furthermore, concrete is a mixture mainly composed of [19] E. Le Bris, Modélisation de la pénétration d’une colle à solvant dans
cement paste and aggregates. The presence of aggregates un milieu poreux, PhD thesis, Ecole Nationale des Ponts et Chaussées,
tends to mask certain features of the thermogravimetric 2003.
curves but does not affect significantly the accuracy of the [20] M.S. Morsy, A.F. Galal, S.A. Abo-El-Enein, Effect of temperature on
phase composition and microstructure of artificial pozzolana-cement
conclusions [4]. Nevertheless, further investigations must pastes containing burnt kaolinite clay, Cem. Concr. Res. 28 (8) (1998)
be done in order to take into account the decomposition 1157 – 1163.
and degradation of aggregates.